Breast magnetic resonance imaging for surveillance of women with a personal history of breast cancer: outcomes stratified by interval between definitive surgery and surveillance MR imaging.

BACKGROUND: Women with a personal history of breast cancer are at increased risk of future breast cancer events, and may benefit from supplemental screening methods that could enhance early detection of subclinical disease. However, current literature on breast magnetic resonance (MR) imaging surveillance is limited. We investigated outcomes of surveillance breast magnetic resonance (MR) imaging in women with a personal history of breast cancer.

METHODS: We reviewed 1053 consecutive breast MR examinations that were performed for surveillance in 1044 women (median age, 53 years; range, 20-85 years) previously treated for breast cancer between August 2014 and February 2016. All patients had previously received supplemental surveillance with ultrasound. Cancer detection rate (CDR), abnormal interpretation rate and characteristics of MR-detected cancers were assessed, including extramammary cancers. We also calculated the PPV 1 , PPV 3 , sensitivity and specificity for MR-detected intramammary lesions. Performance statistics were stratified by interval following initial surgery.

RESULTS: The CDR for MR-detected cancers was 6.7 per 1000 examinations (7 of 1053) and was 3.8 per 1000 examinations (4 of 1053) for intramammary cancers. The overall abnormal interpretation rate was 8.0%, and the abnormal interpretation rate for intramammary lesions was 7.2%. The PPV1 , PPV3 , sensitivity and specificity for intramammary lesions was 5.3% (4 of 76), 15.8% (3 of 19), 75.0% (3 of 4) and 98.3% (1031 of 1049), respectively. For MR examinations performed ≤36 months after surgery, the overall CDR was 1.4 per 1000 examinations. For MR examinations performed > 36 months after surgery, the overall CDR was 17.4 per 1000 examinations.

CONCLUSIONS: Surveillance breast MR imaging may be considered in women with a history of breast cancer, considering the low abnormal interpretation rate and its high specificity. However, the cancer detection rate was low and implementation may be more effective after more than 3 years after surgery.